Following the successful rollout across its journals and books of its AI tool that detects cases of AI-generated nonsense text in research manuscripts, Springer Nature is now donating the tool to The International Association of Scientific, Technical and Medical Publishers (STM), a nonprofit foundation dedicated to advancing trusted research worldwide. It will be integrated into the STM Integrity Hub, an industry-wide initiative that supports publishers in ensuring the integrity of their published content, as part of its mission to develop and trial tools that publishers large and small can use to screen submissions for indicators of compromised content.
Chris Graf, Director of Research Integrity at Springer Nature and Chair of the STM Integrity Hub Governance Committee, commented:
“Developing this tool has been a major investment and a long-running project involving close collaboration between leading research integrity and AI teams. We are delighted we will now be able to share this technology with the wider publishing community so it can have an even bigger impact. The rise of AI has made it easier for unethical individuals to generate fake content and tools like this one which harness the power of AI and pattern recognition will be vital, particularly when they are powered by the extensive data that utilisation across the industry will deliver.”
Dr Joris van Rossum, Program Director of STM Solutions, said:
“The growth in fraudulent submissions from paper mills, facilitated by the rise in generative AI, is an increasing challenge for the publishing community. The STM Integrity Hub is a major initiative to help publishers to combat the scourge of papermills and other unethical conduct. We are delighted to be able to integrate this unique tool that identifies indicators of AI generated manuscripts, to support publishers across the industry. This wider use will help to further train the tool and improve its accuracy.”
The tool has already been responsible for identifying hundreds of fake papers soon after submission, preventing them from being published and taking up editors’ and peer reviewers’ valuable time. in addition, the tool provides a gateway to identifying a larger cohort of problematic submissions. Connections between the original papers and other content, for example as part of the same special issue, can be found, leading to the identification of papers that, at first sight, appear robust but, upon more rigorous analysis, are in fact problematic. Data from all these papers feed into Springer Nature’s wider investigations and help to keep out compromised content.
The nonsense text detector is one of a number of AI tools, including Snappshot, which identifies problematic images, and an irrelevant reference checker tool, which was launched last month, that have been or are being developed in-house as part of Springer Nature’s ongoing commitment to ensuring the integrity of the content it publishes. This commitment also includes investment in a rapidly growing, expert team and extensive technology development and utilisation. In addition, Springer Nature is also involved in ongoing collaborative work with the wider publishing community both as a contributing organisation in the STM Integrity Hub, where, in addition to this donation, it is also involved in multiple working groups and pilot tests, and through involvement with other sector bodies and organisations.
